This is nowhere near as common as you seem to think, and mostly only happens for the narrow cases where somebody is obsessed with a particular problem so that they'd actually want to read other people's solutions. Most of the implementers do not have that sort of relationship to a problem they solved in the past.
If in December you make a general purpose stable sort that's 25% faster than his, Orson Peters is probably going to read your code and try to apply ideas from it. But sorting is the thing Peters really cares about, the people who wrote the stable sort in say Microsoft's STL (their C++ standard library implementation) even if they still work there won't care enough to go back and change that unless told to do so.
If in December you make a general purpose stable sort that's 25% faster than his, Orson Peters is probably going to read your code and try to apply ideas from it. But sorting is the thing Peters really cares about, the people who wrote the stable sort in say Microsoft's STL (their C++ standard library implementation) even if they still work there won't care enough to go back and change that unless told to do so.